Game Theoretic Modeling of Vehicle Interactions at Unsignalized Intersections and Application to Autonomous Vehicle Control
DOI: 10.23919/acc.2018.8430842
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the challenge of modeling vehicle interactions at unsignalized intersections, where the absence of traffic lights requires drivers to make interactive, time-extended decisions to avoid deadlocks and collisions. The authors argue that traditional reachability-based approaches are overly conservative because they assume worst-case scenarios, while existing game-theoretic models often ignore vehicle dynamics or limit interactions to single-step decisions. To address these limitations, the study proposes a game-theoretic framework based on level-k reasoning to model the strategic decision-making of human drivers and applies this model to design a controller for autonomous vehicles. The methodology employs a discrete-time dynamic model for vehicle motion, where agents select from a finite set of maneuvers (e.g., accelerate, brake, turn) to maximize a cumulative reward over a receding horizon. The reward function penalizes collisions, unsafe distances, off-road driving, and violations of lane markings, while encouraging progress toward an objective lane. The core of the interaction model is level-k game theory, which assumes agents have varying depths of strategic reasoning. A level-0 agent treats others as stationary obstacles; a level-1 agent assumes others are level-0; and a level-k agent assumes others are level-(k-1). The autonomous vehicle controller utilizes a multi-model strategy, maintaining a probability distribution over the reasoning levels of interacting drivers. It updates this belief by comparing predicted actions to observed behaviors, allowing it to adapt its strategy to the specific driver it is interacting with. Simulation results evaluate the model in two scenarios involving two vehicles approaching an intersection. In the first scenario, vehicles with mismatched reasoning levels (e.g., level-1 vs. level-0) successfully resolved conflicts, whereas vehicles with identical levels (e.g., level-0 vs. level-0 or level-2 vs. level-2) frequently collided. In a second scenario with randomized initial conditions, conflict resolution rates varied significantly by interaction type: level-1 vs. level-0 achieved a 99% success rate, while level-0 vs. level-0 achieved only 41%. The autonomous controller demonstrated effective model identification, rapidly increasing the probability weight of the correct driver model during interactions. This allowed the controller to select optimal actions that resolved conflicts efficiently, even when interacting with drivers of different reasoning levels. The significance of this work lies in providing a computationally tractable method for modeling complex, interactive human driving behaviors at unsignalized intersections. By incorporating hierarchical reasoning and adaptive belief updates, the proposed controller reduces the conservatism of safety-critical approaches and improves the ability of autonomous vehicles to negotiate right-of-way in unmanaged traffic environments. The findings suggest that accurately predicting the reasoning level of other agents is crucial for successful conflict resolution, offering a pathway for more natural and efficient autonomous driving in urban settings.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | OpenAlex-citations | — | — | 1 | 2026-06-18 |
| archive | success | unpaywall | — | — | 2 | 2026-06-25 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-18 |
| chunk | success | chunk | — | — | 1 | 2026-06-18 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-18 |
| promote | success | — | — | — | 1 | 2026-06-18 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-18 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Theoretical Contribution: computational model